Page MenuHomePhabricator

[X86] PR26575: Fix LEA optimization pass (Part 2)
ClosedPublic

Authored by aturetsk on Feb 18 2016, 4:10 AM.

Details

Summary

Handle address displacement operands of a type other than Immediate or Global in LEAs and load/stores.

Diff Detail

Repository
rL LLVM

Event Timeline

aturetsk updated this revision to Diff 48299.Feb 18 2016, 4:10 AM
aturetsk retitled this revision from to [X86] PR26575: Fix LEA optimization pass (Part 2).
aturetsk updated this object.
aturetsk added reviewers: qcolombet, nadav, hans.
aturetsk added a subscriber: llvm-commits.
hans edited edge metadata.Feb 18 2016, 5:53 PM

I don't know this well enough to review it, but I tried the patch on a full build of Chromium and all its tests on Windows, and everything seemed to build fine.

qcolombet accepted this revision.Feb 19 2016, 9:37 AM
qcolombet edited edge metadata.

Hi Andrey,

LGTM modulo few nitpicks.

Thanks,
Q.

test/CodeGen/X86/lea-opt-memop-check.ll
40 ↗(On Diff #48299)

Could you add a comment on what this test case is checking?

56 ↗(On Diff #48299)

Ditto.

This revision is now accepted and ready to land.Feb 19 2016, 9:37 AM
This revision was automatically updated to reflect the committed changes.

Hello Quentin,

Thanks for the review.
The comments were added in the test before the commit.